{"id":78818,"date":"2025-08-19T23:02:24","date_gmt":"2025-08-19T23:02:24","guid":{"rendered":"https:\/\/www.newsbeep.com\/uk\/78818\/"},"modified":"2025-08-19T23:02:24","modified_gmt":"2025-08-19T23:02:24","slug":"ai-systems-often-prefer-ai-written-content-study-finds","status":"publish","type":"post","link":"https:\/\/www.newsbeep.com\/uk\/78818\/","title":{"rendered":"AI Systems Often Prefer AI-Written Content, Study Finds"},"content":{"rendered":"<p>A peer-reviewed <a href=\"https:\/\/www.pnas.org\/doi\/10.1073\/pnas.2415697122\" target=\"_blank\" rel=\"noopener nofollow\">PNAS study<\/a> finds that large language models tend to prefer content written by other LLMs when asked to choose between comparable options.<\/p>\n<p>The authors say this pattern could give AI-assisted content an advantage as more product discovery and recommendations flow through AI systems.<\/p>\n<p>About The Study<br \/>\nWhat the researchers tested<\/p>\n<p>A team led by Walter Laurito and Jan Kulveit compared human-written and AI-written versions of the same items across three categories: marketplace product descriptions, scientific paper abstracts, and movie plot summaries.<\/p>\n<p>Popular models, including GPT-3.5, GPT-4-1106, Llama-3.1-70B, Mixtral-8x22B, and Qwen2.5-72B, acted as selectors in pairwise prompts that forced a single pick.<\/p>\n<p>The paper states:<\/p>\n<p>\u201cOur results show a consistent tendency for LLM-based AIs to prefer LLM-presented options. This suggests the possibility of future AI systems implicitly discriminating against humans as a class, giving AI agents and AI-assisted humans an unfair advantage.\u201d<\/p>\n<p>Key results at a glance<\/p>\n<p>When GPT-4 provided the AI-written versions used in comparisons, selectors chose the AI text more often than human raters did:<\/p>\n<p>Products: 89% AI preference by LLMs vs 36% by humans<br \/>\nPaper abstracts: 78% vs 61%<br \/>\nMovie summaries: 70% vs 58%<\/p>\n<p>The authors also note order effects. Some models showed a tendency to pick the first option, which the study tried to reduce by swapping the order and averaging results.<\/p>\n<p>Why This Matters<\/p>\n<p>If marketplaces, chat assistants, or search experiences use LLMs to score or summarize listings, AI-assisted copy may be more likely to be selected in those systems.<\/p>\n<p>The authors describe a potential \u201cgate tax,\u201d where businesses feel compelled to pay for AI writing tools to avoid being down-selected by AI evaluators. This is a marketing operations question as much as a creative one.<\/p>\n<p>Limits &amp; Questions<\/p>\n<p>The human baseline in this study is small (13 research assistants) and preliminary, and pairwise choices don\u2019t measure sales impact.<\/p>\n<p>Findings may vary by prompt design, model version, domain, and text length. The mechanism behind the preference is still unclear, and the authors call for follow-up work on stylometry and mitigation techniques.<\/p>\n<p>Looking ahead<\/p>\n<p>If AI-mediated ranking continues to expand in commerce and content discovery, it is reasonable to consider AI assistance where it directly affects visibility.<\/p>\n<p>Treat this as an experimentation lane rather than a blanket rule. Keep human writers in the loop for tone and claims, and validate with customer outcomes.<\/p>\n","protected":false},"excerpt":{"rendered":"A peer-reviewed PNAS study finds that large language models tend to prefer content written by other LLMs when&hellip;\n","protected":false},"author":2,"featured_media":78819,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[20],"tags":[554,733,4308,86,56,54,55],"class_list":["post-78818","post","type-post","status-publish","format-standard","has-post-thumbnail","category-artificial-intelligence","tag-ai","tag-artificial-intelligence","tag-artificialintelligence","tag-technology","tag-uk","tag-united-kingdom","tag-unitedkingdom"],"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/posts\/78818","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/comments?post=78818"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/posts\/78818\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/media\/78819"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/media?parent=78818"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/categories?post=78818"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/tags?post=78818"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}